AI Paper Atlas

Research map / Adversarial robustness and security / Attacks on text and graphs

DIFFender: Diffusion-Based Adversarial Defense against Patch Attacks

Caixin Kang, Yinpeng Dong, Zhengyi Wang, Shouwei Ruan, Yubo Chen, Hang Su, Xingxing Wei

ECCV 2024 · Adversarial robustness and security › Attacks on text and graphs · approach: Diffusion-based adversarial purification · 28 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add cross-entropy mask loss L_CE to prompt tuning (ablation)DIFFender · ImageNet BPDA+AdvProbust accuracy (%): 76.2 -> 88.3Improved
add L1 restoration loss to prompt tuning (ablation)DIFFender · ImageNet cleanaccuracy (%): 88.3 -> 91.4Improved
add feature-distance loss d to prompt tuning (ablation)DIFFender · ImageNet BPDA+AdvProbust accuracy (%): 87.1 -> 88.3Improved
add patch restoration after localization (vs. zeroing masked region) (ablation)DIFFender · ImageNet BPDA+AdvProbust accuracy (%): 84.0 -> 88.3Improved
use few-shot tuned prompts instead of empty prompt (ablation)DIFFender · ImageNet BPDA+AdvProbust accuracy (%): 76.4 -> 88.3Improved
use few-shot tuned prompts instead of manual prompts (ablation)DIFFender · ImageNet BPDA+AdvProbust accuracy (%): 77.9 -> 88.3Improved
use plain DiffPure purification against patch attacksInception-v3 · ImageNet BPDA+AdvProbust accuracy (%): 10.5Hurt
tune prompts on Inception-v3 only, transfer to other classifiersDIFFender · ImageNet ResNet-50 AdvProbust accuracy (%): Jedi 46.9 -> DIFFender 83.2Improved
use DIFFender vs Jedi/SAC on larger patchesDIFFender · ImageNet AdvP 15% patchrobust accuracy (%): SAC 34.8 -> DIFFender 56.6Improved
apply feature-norm clipping defense to a transformer classifierFNC · ImageNet BPDA+AdvProbust accuracy (%): Inception-v3 61.3 -> Swin-S 6.3Hurt

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy